Delay Aware Packet Scheduling (DAPS) and receivers buffer blocking - - PowerPoint PPT Presentation

▶

Dec 12, 2022 412 likes •560 views

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion Delay Aware Packet Scheduling (DAPS) and receivers buffer blocking in CMT-SCTP Nicolas KUHN 1 , 2 , Golam SARWAR 2 , Emmanuel LOCHIN 1 , Roksana BORELI

SLIDE 1

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

Delay Aware Packet Scheduling (DAPS) and receiver’s buffer blocking in CMT-SCTP

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1 ISAE / T´

eSA (Toulouse, France)

2 National ICT Australia

nicolas.kuhn@isae.fr

July 31, 2013

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 2

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

Context and problem definition

Context: many recent studies on ways to adapt TCP to the multi-path data transfert; we focus on Concurrent Multipath Transfer Using SCTP Multihoming (solution proposed may be transfered to MP-TCP); asymmetric heterogeneous network conditions: receiver’s buffer blocking issues (out of order packet arrival). Content of the presentation: measure the receiver’s buffer blocking time; propose a solution (adaptation of CMT-SCTP scheduler) for that specific issue; compare CMT-SCTP with and without our scheduler, Delay Aware Packet Scheduler.

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 3

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

Overview

1 Window blocking issues and maximum blocking time 2 Delay Aware Packet Scheduling 3 Conclusion

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 4

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

Maximum blocking time with CMT-SCTP

t0 t2 t1 RTTm/2 RTTM/2 5 4 3 2 1

Events

at t0: (1) TSN(2) . . . TSN(Nchunk) on path mRTT - (2) TSN(1) on path MRTT; at t0: RWND: block state; at t1: reception of TSN(2); at t2: reception of TSN(1) - packet forwarded to app.;

Tmaxblock = t2 − t1 Tmaxblock = RTTMRTT

2

+

L∗8 CMRTT − RTTmRTT 2

−

L∗8 CmRTT

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 5

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

Validation blocking time

Worst case model

C0 C1 RTT1 Tmaxblock (Mb/s) (Mb/s) (ms) Model 10 10 100 0.040 50 10 100 0.041 10 50 100 0.039 100 100 100 0.040 50 150 100 0.039 150 50 100 0.040 10 10 200 0.090 50 10 200 0.091 10 50 200 0.089 100 100 200 0.090 50 150 200 0.089 150 50 200 0.090

NS-2

0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 150 160 170 180 190 200 Blocking time (s) Packet ID C0 100Mbps, RTT1 100ms,C1 100Mbps (NS−2) C0 100Mbps, RTT1 100ms,C1 100Mbps (MODEL) C0 100Mbps, RTT1 200ms,C1 100Mbps (NS−2) C0 100Mbps, RTT1 200ms,C1 100Mbps (MODEL) C0 10Mbps, RTT1 200ms,C1 10Mbps (NS−2) C0 10Mbps, RTT1 200ms,C1 10Mbps (MODEL)

We measure an important blocking time in the context of asymmetric links. We propose to adapt the scheduling in CMT to reduce this delay which may “block” the receiver’s buffer.

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 6

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

Overview

1 Window blocking issues and maximum blocking time 2 Delay Aware Packet Scheduling 3 Conclusion

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 7

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

Delay Aware Packet Scheduling

CMT

1 2 3 4 5 6 7 8 9 10 BLOCK! RWND=10 deliver 1 to app deliver 2, ..., 10 to app

n path 1
n path 2

DAPS

RWND=10 2 1 3 4 5 6 7 8 9 10 11 12 13 14 15

n path 1
n path 2

DAPS: compute {14,15} to send on low path for in sequence delivery Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 8

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

DAPS performance

Butterfly

1 2 3 4 5 RTT0, C0 RTT1, C1 RWND

Parameters RWND = 655kB; L = 1500B; RTT0 = 20ms; C0 = [1Mbps; 1.5Mbps]; RTT1 ∈ [100; 200]ms; C1 ∈ [500kbps; 1Mbps] Asymmetry is defined as: A = (C1.RTT1)/(C0.RTT0)

Case C0 C1 RTT1 A (Mb/s) (Mb/s) (ms) 1 1 1 100 5 2 1 1 200 10 3 1.5 0.5 200 3.33

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 9

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

Asymmetry: 5 DAPS Case 1

200 400 600 800 1000 1200 1400 1600 1800 2 4 6 8 10 12 Packet ID Simulation time (s) Packet in Sending Buffer Packet received in order 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 200 400 600 800 1000 1200 1400 1600 1800 Blocking time (s) Packet ID

CMT Case 1

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 10

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

Asymmetry: 10 DAPS Case 2

200 400 600 800 1000 1200 1400 1600 1800 2 4 6 8 10 12 Packet ID Simulation time (s) Packet in Sending Buffer Packet received in order 0.1 0.2 0.3 0.4 0.5 0.6 200 400 600 800 1000 1200 1400 1600 Blocking time (s) Packet ID

CMT Case 2

200 400 600 800 1000 1200 1400 1600 1 2 3 4 5 6 7 8 9 10 Packet ID Simulation time (s) Packet in Sending Buffer Packet received in order 0.1 0.2 0.3 0.4 0.5 0.6 0.7 200 400 600 800 1000 1200 1400 1600 Blocking time (s) Packet ID

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 11

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

Asymmetry: 3.33 DAPS Case 3

200 400 600 800 1000 1200 1400 1600 1800 2 4 6 8 10 12 Packet ID Simulation time (s) Packet in Sending Buffer Packet received in order 0.2 0.4 0.6 0.8 1 1.2 1.4 200 400 600 800 1000 1200 1400 1600 Blocking time (s) Packet ID

CMT Case 3

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 12

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

Overview

1 Window blocking issues and maximum blocking time 2 Delay Aware Packet Scheduling 3 Conclusion

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)

SLIDE 13

Window blocking issues and maximum blocking time Delay Aware Packet Scheduling Conclusion

DAPS Performance: first conclusions

In case 2 (Asymmetry: 10), DAPS provide serious improvements; In case 1 (Asymmetry: 5), improvements are limited (without doing worse than CMT-SCTP); In case 3 (Asymmetry: 3.33), improvements are limited. The performance of DAPS are related to: the Asymmetry of the links; the receiver’s buffer, which we will later investigated on to improve the algorithm.

Nicolas KUHN1,2, Golam SARWAR2, Emmanuel LOCHIN1, Roksana BORELI2, Ahlem MIFDAOUI1

1ISAE / T´

eSA, 2NICTA Delay Aware Packet Scheduling (DAPS)